Emotion classification method based on feature selection and feature extraction

The invention discloses an emotion classification method based on feature selection and feature extraction, and the method is specifically implemented according to the following steps: obtaining an English text corpus, carrying out the preprocessing of the English text corpus, obtaining an English text data set, carrying out the word segmentation of all texts in the English text data set, and obtaining a word sequence; obtaining feature words in the word sequence by adopting a syntactic dependency relationship and part-of-speech features, and forming a feature word set by the feature words; performing feature extraction on each feature word in the feature word set by adopting an improved TF-IDF algorithm to obtain a weight representation of each feature word, forming a weight vector of the feature word set by the weight representation of each feature word, and forming a weight matrix through the weight vectors; performing weighted fusion on the weight matrix and an attention matrix output by the BERT model to